Research map / 3D vision / NeRF and radiance fields
3D-aware generative radiance fields
Trains GAN-based radiance fields to generate 3D-consistent images.
20 accepted papers at ICML, NeurIPS, ICLR, CVPR and AAAI (2020–2026) · an approach in NeRF and radiance fields, keyword cluster “rendering · field · synthesis”
State History Benchmarks Tried Baselines Preprints Who Export
Where it stands
Quiet None of the tracked benchmarks has a report from 2025–26.
11 papers in 2023–24 and 1 in 2025–26 (at ICML, NeurIPS, ICLR, CVPR and AAAI); its share of the topic shrank, from 4.1% to 1.1%.
Open questions the data points to
- Papers report 2D GAN on Cars, 64² (FID) anywhere from 16 to 66: comparisons against it depend on whose number you take.
- Papers report GRAF on FFHQ, 256² (FID) anywhere from 59 to 73: comparisons against it depend on whose number you take.
No longer reported by papers here: FFHQ 256² (FID, last in 2022).
How this is decided
A benchmark is tracked when at least four papers of this approach report their own method on the same dataset and metric in at least two different years. Its record is the best number reported up to 2024 compared with the best reported in 2025–26: still improving means at least 1% better, flat means at most 0.2% better with at least two papers from 2025–26 trying. Numbers on a different scale from the rest (a fraction among percentages) and far outliers are left out. Results come from different papers with different setups, so they are signals, not head-to-head comparisons. The share of papers is counted at ICML, NeurIPS, ICLR, CVPR and AAAI.
How it got here
Most cited paper of each year
- 2020 · GRAF: Generative Radiance Fields for 3D-Aware Image Synthesis (NeurIPS 2020 · 319 citations)
- 2021 · GIRAFFE: Representing Scenes As Compositional Generative Neural Feature Fields (CVPR 2021 · 742 citations)
- 2022 · GRAM: Generative Radiance Manifolds for 3D-Aware Image Generation (CVPR 2022 · 202 citations)
- 2023 · One-Shot High-Fidelity Talking-Head Synthesis With Deformable Neural Radiance Field (CVPR 2023 · 54 citations)
- 2024 · DL3DV-10K: A Large-Scale Scene Dataset for Deep Learning-based 3D Vision (CVPR 2024 · 140 citations)
- 2026 · Scaling View Synthesis Transformers (CVPR 2026 · 0 citations)
Earliest papers here
- GRAF: Generative Radiance Fields for 3D-Aware Image Synthesis (NeurIPS 2020 · 319 citations)
- GIRAFFE: Representing Scenes As Compositional Generative Neural Feature Fields (CVPR 2021 · 742 citations)
- Pi-GAN: Periodic Implicit Generative Adversarial Networks for 3D-Aware Image Synthesis (CVPR 2021 · 698 citations)
Benchmark records
The best number each year that beat every earlier one, among papers of this approach reporting their own method. Numbers are extracted by AI from each paper's text; check the paper before citing.
FFHQ 256² · FID (lower is better)
4 papers, 2021–2022 · not reported since 2022
| Year | Best so far | Set by |
|---|---|---|
| 2021 | 32 | GIRAFFE: Representing Scenes As Compositional Generative Neural Feature Fields (CVPR 2021 · 742 citations) |
| 2022 | 8 | StyleNeRF: A Style-based 3D Aware Generator for High-resolution Image Synthesis (ICLR 2022 · 179 citations) |
What has been tried
18 papers here report 196 measured changes (a component added, removed or swapped, compared with a baseline): helped 73%, no clear effect 7%, hurt 19%, mixed 1%. Papers mostly publish changes that helped, so "helped" is over-represented.
Ablations in recent papers
| What was changed | Measured on | Result | Effect | Paper |
|---|---|---|---|---|
| Use Zip-NeRF's default batch size instead of 4,096 ray samples per batch | DL3DV-140 · PSNR | 29.07 -> 31.22 | Helped | DL3DV-10K: A Large-Scale Scene Dataset for Deep Learning-based 3D Vision (CVPR 2024 · 140 citations) |
| add synthetic multi-view data | FFHQ · Depth | 0.42 -> 0.35 | Helped | G-NeRF: Geometry-enhanced Novel View Synthesis from Single-View Images (CVPR 2024 · 6 citations) |
| Add Gaussian timestep sampling (mean = 1000, std=200) | Objaverse validation · LPIPS | 0.236 -> 0.205 | Helped | Efficient-3Dim: Learning a Generalizable Single-image Novel-view Synthesizer in One Day (ICLR 2024 · 2 citations) |
Baseline numbers that disagree
The same baseline on the same benchmark and metric, reported more than 5% apart by different papers here. Training budgets, data splits and reimplementations differ, so state which number you compare with.
| Baseline | Benchmark | Metric | Reported | Papers |
|---|---|---|---|---|
| 2D GAN | Cars, 64² | FID | 16 – 66 | 2 papers, e.g. CVPR 2021 and NeurIPS 2020 |
| GRAF | FFHQ, 256² | FID | 59 – 73 | 2 papers, e.g. CVPR 2021 and CVPR 2022 |
| GIRAFFE | FFHQ 256² | FID | 32.6 – 35 | 2 papers, e.g. CVPR 2022 and ICLR 2022 |
Recent preprints
Loading recent arXiv preprints…
Who publishes on it
Institutions on the most papers of this approach since 2023 (of 12 papers; from OpenAlex and the papers' affiliations).
- Zhejiang University · 3 papers
- ByteDance · 2 papers
- Ant Group · 2 papers
- The Chinese University of Hong Kong · 2 papers
Use it
Download BibTeX of 20 key papersOpen the topic in the interactive map
Get alerts. Describe your idea in scime atlas and save it (free scime account): you get a weekly email when new preprints land close to it.
Related approaches
Same keyword cluster
- Inverse rendering with neural fields (110 papers)
- Few-shot neural rendering architectures (53 papers)
Elsewhere in the topic
- Sparse-input NeRF regularization (106 papers)
- Dynamic and editable NeRF (88 papers)
- Single-view and streaming radiance fields (49 papers)
- NeRF for scene decomposition and understanding (30 papers)
Benchmark numbers, ablations and baseline numbers are extracted by AI from each paper's full text and can be wrong; the paper is the authority. Data: results index of 2026-11-02, citation counts from OpenAlex (2026-10-09).